Skip to main content

Delve: AI-Powered Taxonomy Generation

Delve is a production-ready SDK and CLI for automatically generating taxonomies from your data using state-of-the-art language models.

📚 Read the full documentation →

Quick Start

Installation

pip install delve-taxonomy

# Set API keys
export ANTHROPIC_API_KEY="your-key-here"
export OPENAI_API_KEY="your-key-here"  # Required for classifier embeddings

CLI

# Basic usage (shows progress spinners)
delve run data.csv --text-column text

# With progress bars and ETA
delve run data.csv --text-column text -v

# Quiet mode (errors only)
delve run data.csv --text-column text -q

# JSON with nested data
delve run data.json --json-path "$.messages[*].content"

Python SDK

from delve import Delve, Verbosity

# Initialize client (silent by default - library best practice)
delve = Delve()

# Or with progress output
delve = Delve(verbosity=Verbosity.NORMAL)

# Run taxonomy generation
result = delve.run_sync("data.csv", text_column="text")

# Access results
print(f"Generated {len(result.taxonomy)} categories")
for category in result.taxonomy:
    print(f"  - {category.name}: {category.description}")

# Access labeled documents
for doc in result.labeled_documents[:5]:
    print(f"  [{doc.category}] {doc.content[:50]}...")

Binary Detection (Single Category)

For fast filtering when you know the category you're looking for:

from delve import Delve

# Find all refund-related documents (~$1-2 for 30K docs, runs in minutes)
result = Delve.find_matches(
    "data.csv",
    category={
        "name": "Refund Request",
        "description": "User asking for refund or money back",
        "keywords": ["refund", "money back", "cancel"],
    },
    text_column="text",
    threshold=0.6,
)

print(f"Found {result.stats['matches']} matches")
for doc in result.matched_documents[:5]:
    print(f"  [{doc.confidence:.2f}] {doc.content[:50]}...")

Features

  • Automated Taxonomy Generation - No manual category creation — powered by the latest Claude models
  • Binary Detection - Fast, cheap single-category filtering with find_matches()
  • Multiple Data Sources - CSV, JSON/JSONL, LangSmith runs, pandas DataFrames
  • Smart Categorization - Iterative refinement with minibatch clustering
  • Flexible Exports - JSON, CSV, and Markdown reports

Requirements

  • Python 3.9+
  • Anthropic API key (for taxonomy generation)
  • OpenAI API key (for classifier embeddings when sample_size > 0)

Documentation

Development

# Install dependencies
uv sync

# Run tests
pytest tests/

# Run linting
ruff check src/

# Format code
ruff format src/

Documentation Development

To work on the documentation locally, you'll need Node.js 20.17+ (for Mintlify):

# If using nvm, the project includes .nvmrc
nvm use

# Install Mintlify CLI (if not already installed)
npm install -g mintlify

# Run the docs server
cd docs
mintlify dev

See the full documentation for more details on contributing and development.

Metadata

Release files for delve-taxonomy 0.2.1

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for delve-taxonomy 0.2.1
File Size Uploaded Provenance Actions
delve_taxonomy-0.2.1.tar.gz 54.7 kB With provenance Details for delve_taxonomy-0.2.1.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for delve-taxonomy 0.2.1
File Interpreter ABI Platform Actions
delve_taxonomy-0.2.1-py3-none-any.whl Python 3 none any Details for delve_taxonomy-0.2.1-py3-none-any.whl

Total release size: 122.8 kB

Release files / delve_taxonomy-0.2.1.tar.gz

Download URL delve_taxonomy-0.2.1.tar.gz
Size 54.7 kB
Tags Source
SHA-256 checksum
How to use checksums
b5a0fc25abeb9159dd7a17bfa4c3178689ee488ea384fd3b69b3e440fbc895a0
BLAKE2b-256 checksum
How to use checksums
343c5b253b7918a788d0aef1c5420a70e620bc51769aa52aa566b8f72fd7ae84
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 16, 2026.

Transparency log

Release files / delve_taxonomy-0.2.1-py3-none-any.whl

Download URL delve_taxonomy-0.2.1-py3-none-any.whl
Size 68.1 kB
Tags Python 3
SHA-256 checksum
How to use checksums
7a8d043f250b0fb8184e7256544e106368942c27e78159a07df97e7f785e6dbe
BLAKE2b-256 checksum
How to use checksums
8cf912905acd76b4dfa3f8ccecb6c2fd0ebf671c83ce0e8fdc8ef0cd11d235f9
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 16, 2026.

Transparency log

Security for delve-taxonomy 0.2.1

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

Full provenance

Both files in this release have been uploaded with provenance metadata.

Select from 2 files

Attestation for delve_taxonomy-0.2.1.tar.gz

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 16, 2026.

Transparency log

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page